США ужесточили надзор за LLM, а ИИ расшифровал обугленный свиток целиком

In the US, new mandatory requirements have been introduced for developers of the most powerful language models — companies must notify regulators before publicly releasing a system. In parallel, an international team of scientists using neural networks has for the first time fully deciphered a charred Herculaneum papyrus — a scroll that lay untouched for nearly two millennia.

What Changed for LLM Developers

American regulators have tightened oversight of companies building the most powerful language models. New rules require major players to notify authorities before releasing particularly capable systems and to conduct dual-use risk assessments — including in military and biological contexts.

Key requirements of the new regime:

• Mandatory notification of regulators prior to public model release
• Assessment of dual-use potential: cyberattacks, biological threats
• Provision of technical documentation upon request by authorities
• Measures are aimed primarily at OpenAI, Anthropic, Google DeepMind, and Meta

According to official statements from the agencies, the requirements are not designed to slow innovation — the goal is to create an early-warning system in case models with unforeseen dangerous capabilities emerge.

Just a few years ago, the US relied primarily on voluntary agreements with AI labs: in July 2023, major companies signed "voluntary commitments" on safety under the auspices of the White House. The new regime establishes legal — not merely reputational — obligations for frontier technology developers. Similar rules have already come into force in the European Union under the EU AI Act: since February 2025, AI practices with an unacceptable level of risk have been banned in the EU, and requirements for general-purpose models became mandatory in August 2025.

How a Neural Network Read a Papyrus After 2,000 Years

Researchers in the open competition Vesuvius Challenge announced the complete reading of one of the Herculaneum papyri. These scrolls were charred during the eruption of Vesuvius in 79 AD and are kept at the National Library of Naples. Before AI, physical unrolling inevitably destroyed the fragile documents — which is why most scrolls remained unreadable for about two thousand years.

The team applied a virtual unrolling method: an X-ray tomograph creates layered three-dimensional scans of the rolled scroll, and a computer vision neural network analyzes the minutest contrast differences between the ink and the charred surface of the papyrus. The model "sees" the letters through the layers of the coiled document — without a single physical intervention in the original.

"For the first time in nearly 2,000 years, we are reading a text that the last person held in their hands shortly before the destruction of

Herculaneum," the organizers of Vesuvius Challenge stated in the project's official blog.

The deciphered scroll contains a philosophical text presumably of the Epicurean tradition — attributed to Philodemus of Gadara. The Herculaneum library is the only one to have survived from antiquity: approximately 1,800 scrolls were discovered underground, the majority still undeciphered. Among them may be lost works of Epicurus, Aristotle, and other philosophers, of whom only fragments have reached us. Vesuvius Challenge was founded in 2023 with a prize fund of more than one million dollars.

What This Means

Both events are not a coincidental pairing in the news feed. They show the maturity of AI from different angles: the technology is powerful enough to attract the attention of state regulators, and at the same time precise enough to recover texts that had been silent for nearly two millennia. Regulation and scientific discovery are accelerating in parallel — and both processes are powered by the same language models.

*Meta has been recognized as an extremist organization and is banned in Russia.